The present invention relates to a reticle and a method of fabricating a reticle.
The semiconductor integrated circuit has been steadily increase in their high integration and miniaturization properties. In the fabrication of the highly integrated semiconductor circuit, a lithography technique plays a vital role.
Most of the lithography techniques now employ a master pattern which is projected on an LSI substrate through a reducing optical system. In these techniques, however, when a high voltage mercury lamp is employed as a light source, the minimum linewidth is about 0.5 .mu.m. For the purpose of obtaining a pattern size of less than 0.5 .mu.m, there have been developed a direct writing technique using a KrF excimer laser or an electron beam, and an X-ray unity magnification lithography technique. However, from the viewpoint of mass productivity, process versatility and so on, great expectation has been directed to the photolithography technique.
Under such circumstance, various studies have been made in the photolithography technique. For a light source, g ray, i ray, X ray, excimer laser, etc. have been studied. For a resist, possibility of development of new resist materials and new resist processing such as REL have been studied. Further SREP.CEL image reverse technique has been studied.
With respect to the mask fabrication technique, sufficient studies had not been made until Levenson, et al. of IBM Corporation have proposed a phase shift method in 1982. Since then, attention has been directed to the phase shift method. In the phase shift method, the phase of a beam passing through a mask is controlled for the purpose of enhancing the resolution and contrast of a projected image.
The principle of this method will be explained by referring to FIG. 20(a)-(d). As shown in FIG. 20(a), a mask comprises a mask pattern 12 of laminated films made of chrome (Cr) or chrome oxide (Cr.sub.2 O.sub.3) formed on a quartz substrate 11 by a sputtering or the like process, and a transparent film 13 formed on one of an adjacent pair of light transmissible areas in the mask pattern 12. The phase of light passing through the transparent film area is inverted as shown in FIG. 20(b) and light passed through the adjacent transmissible areas are canceled each other at the boundary part between the two transmissible areas as shown in FIG. 20(c). As a result, light intensity at the boundary part between the two light transmissible areas becomes zero and thus the two light transmissible areas can be formed in separate patterns on the wafer as shown in FIG. 20(d). In this method, a pattern of 0.7 .mu.m can be resolved by using a g-ray stepper having a numerical aperture (NA) of 0.28, thus improving the resolution by about 40%. In this method, in order to inverse the phase of the light passing through the transparent film 13, the transparent film 13 as a phase shifter must satisfy a relationship of d=.lambda./2(n-1), where `d` denotes the thickness of the film, `n` denotes the refractive index of the material of the phase shifter, and `.lambda.` denotes the wavelength of light to be radiated.
Terazawa et al. of Hitachi Ltd. have further developed the method based on Levenson et al. and proposed a new technique in which the Levenson technique is applied to an isolated pattern, which principle is shown in FIG. 21(a)-(d). In the Terazawa technique, an isolated pattern `a` and an auxiliary pattern `b` as a dummy which is not resolved by itself are provided, and phase shifters 13 for inverting the light passing therethrough are provided to the patterns. In this technique, an isolated space of 0.3 .mu.and a contact hole having a diameter of 0.4 .mu.are resolved in an i-ray stepper, having a numerical aperture (NA) of 0.42. Thus, the resolution is improved by about 30% compared with that of the prior art technique. However, with respect to the contact hole, as the pattern sizes become smaller, the difference in light intensity between the isolated and auxiliary patterns becomes smaller. Thus, this technique has resolution limit because the intensity of light passing through the transparent areas is weakened as the pattern size becomes smaller.
The aforementioned prior art technique has additional problems that, since the phase shifter is located for every other light transmissible area in a line-and-space pattern while a mask layer for the auxiliary pattern is processed such that the phase shifter is provided for the isolated pattern, the position alignment of the shifters to the mask pattern and a selective processing technique become necessary and the number of necessary steps will be greatly increased and thus the mask fabrication steps becomes complicated.
In view of such problems, Nitayama, et al. of Toshiba Corporation have suggested a phase shift mask structure in which a phase shifter is provided at the periphery of a light transmissible area or a light blocking area. The principle of this structure is shown in FIG. 22(a)-(b). As shown in FIG. 22(a), a mask comprises a mask pattern 12 of chrome (Cr) and chrome oxide (Cr.sub.2 O.sub.3) formed on a quartz substrate 11, and a transparent film of a phase shifter 13 formed as protruded from the periphery of the mask pattern 12. When such a mask is used, the phases of lights passing through the shifters are inverted so that the amplitudes of the lights with phases 0 degree and 100 degree cancel each other at both ends of the light transmissible area as shown in FIG. 22(b). Thus, the light intensity becomes small and the contrast is improved as shown in FIG. 22(c). As a result, the light intensities at the both ends of the light transmissible area become substantially zero and thus the two light transmissible areas can separate the pattern formed on the wafer as shown in FIG. 22(d).
The above mask is formed in the following manner. First, as shown in FIG. 23(a), a laminated film 12 of about 100 nm of chrome (Cr) and chrome oxide (Cr.sub.2 O.sub.3) is deposited on the quartz substrate 11 by sputtering or like processes. A resist coated on the film 12 is subjected to electron beam writing operation, and then developed so as to form a resist pattern R.
Next, as shown in FIG. 23(b), the resultant substrate is subjected to an etching operation by wet etching or by reactive ion etching process with use of the resist pattern as a mask to remove the resist pattern R and form the mask layer 12. Thereafter, as shown in FIG. 23(c), the resultant substrate is coated with a PMMA film, and is exposed with light from the back face to form a latent image 13S. Then the resultant substrate is subjected to a developing operation to form phase shifters 13 of a PMMA film pattern as shown in FIG. 23(d). Finally, as shown in FIG. 23(e), the mask layer 12 is subjected to a side etching with use of the phase shifters 13 of the PMMA film pattern as a mask, whereby such a mask that the phase shifter 13 is protruded from the associated mask layer 12 is fabricated.
In this mask, the PMMA acts as the phase shifter. Since it has a high transmissivity ratio and a sharp resist profile, the PMMA is an excellent phase shifter. According to this method, the pattern of the phase shifter can be formed on a self alignment basis. Therefore, the mask alignment and selective processing steps are not required and thus a mask can be formed relatively easily.
With the mask, the phase of the light passed through the respective shifters is inverted as shown by a broken line in FIG. 22(c), the light intensity behind the mask layer is remarkably reduced so that intensity distribution of the whole light is as shown by a solid line in FIG. 22(d). This mask has a resolution as fine as nearly half that of the conventional mask.
Assuming that the transmissivity ratio of the phase shifter in the mask is 100%, the width of the shifter is optimized. It is found that the optimum shift width causing the greatest contrast of image varies depending on the pattern size. For example, as shown in FIG. 24, when an excimer stepper having a numerical aperture NA of 0.42 is employed, the optimum shift width is 0.04 .mu.m (on the wafer) with respect to 0.3 .mu.m of line-and-space pattern while the optimum shift width is 0.06 .mu.m (on the wafer) with respect to 0.25 .mu.m of line-and-space pattern. In this manner, the optimum shift which realizes the maximum contrast is determined according to the pattern size. The mask having the optimum shift width enables to resolve a fine pattern that can not be resolved in the conventional technique.
However, in fabricating the above mask, it is necessary to provide a fine shifter pattern on the fine pattern, which requires a very difficult work of controlling the shifter width.
Further, the shifter must be made of either a light transmissible film or a light blocking film. Generally, a mask used in the VLSI fabrication must be frequently cleaned because the mask must be used completely free of dust. For this reason, the phase shift mask must be strong enough to withstand the repetitive cleaning. The shifter formed with resist, however, is not strong enough to withstand the frequent cleaning.
As described above, reticles used in the conventional phase shift technique is defective in that it is very difficult to provide the fine shifter pattern on the fine pattern while controlling the shifter width and making alignment. In addition, the mask is not strong enough to withstand the VLSI fabrication processes.
Further, in order to make the most of the effect of the phase shifting, it is important to optimize a phase difference between light passed through the transparent film and light passed through a light semi-transparent film as well as an amplitude transmissivity ratio therebetween. The phase difference and the amplitude transmissivity ratio are uniquely determined by the optical constant (complex refractive index n-ik, where i is unit imaginary number) of these films. In other words, in order to obtain a desired phase difference and a desired amplitude transmissivity ratio, a certain relationship must be satisfied between the optical constant and the film thickness. However, because the optical constant is inherent in the substance, it is difficult to satisfy the desired relationship with use of a single layer film.
FIG. 30 shows a structure of a conventional ideal half-tone phase film. A mask formed according to the conventional method comprises a light transmissible film 301 and a semi-transparent film. The semi-transparent film is formed to have an amplitude transmissivity ratio of10 to 40% with respect to the light transmissible film and to change the phase of light passing through the semi-transparent film by 180 degrees with respect to the light transmissible film. For the purpose of satisfying these conditions, the semi-transparent film has a two-layer structure which comprises a first layer 302 for adjusting the amplitude transmissivity ratio and a second layer 303 for adjusting the total phase difference of the first and second layers 302 and 303 to be 180 degrees.
In the conventional half-tone phase shift method, the half-tone film has a two-layer structure comprising the first layer for adjusting the amplitude transmissivity ratio and the second layers for adjusting the total phase difference generated through the first and second layers to be 180 degrees. However, the two-layer structure is difficult to fabricate, since it requires pattern transfer and the first and second layers must be processed in the same sizes.
Especially, it is extremely difficult to cure a defect when the lower layer 302 becomes defective as shown in FIG. 31.